递归流分类算法研究与改进

Research and Improvement of Recursive Flow Classification Algorithm

流分类算法的性能直接影响防火墙、路由器等设备的处理速度。递归流分类(RFC)算法具有分类速度快的优点,但随着规则数目的增大,存储开销也随之增加。为此,通过对RFC算法进行分析,提出一种改进算法Optimize_RFC,对块的位数进行异或运算,压缩等价类表,减少内存消耗。实验结果表明,Optimize_RFC算法在保持相对较快分类速度的同时,可降低预处理阶段的内存占用。

Data flow classification algorithm performance directly affects the processing speed in firewalls,routers or other equipments.Recursive Flow Classification（RFC）algorithm has the advantage as faster packet classification,but also exists the disadvantage as huge memory consumption with the number of rules increases.By analyzing the shortcomings of RFC algorithm,this paper proposes an improved algorithm called Optimize_RFC optimizing the equivalence class table with bit XOR operation,reducing memory consumption.Experimental results show that Optimize_RFC algorithm can greatly reduce the memory consumption while maintaining a relatively high speed of classification in the preprocessing stage.